Water-level measurements were made in piezometers installed at two clay-rich, field research sites located in Saskatchewan, Canada. One site was a thick, fractured till and the other of a thick, non-fractured till. At the non-fractured till site, water-level measurements were made in 21 piezometers installed from 2 m to 91 m below ground since their installation in 1995. Seasonal variations in the water levels were observed to a depth of 15 m. At the fractured till site, water-level variations were measured in seven piezometers installed from 5 m to 54 m below ground surface since 1998. Seasonal variations were observed to a depth of 23 m. Seasonal fluctuations in water levels can be represented by natural time series, whereby each observation is associated with a particular representation of time or interval of time and, as such, can be numerically examined by harmonic analyses. Each water-level measurement (as a function of time) was decomposed into a sum of trigonometric components and the phase and amplitude of each harmonic function of a certain frequency was calculated. The validity of the decomposition was tested by comparing measured and calculated time series, using up to 60 frequencies. Excellent agreement was achieved. Vertical hydraulic diffusivities were calculated at various depths assuming a one-dimensional transient flow and applying the computed amplitude ratios and phase lags of neighbouring data series. The vertical hydraulic conductivities (K) were calculated using in situ values of specific storage of till. These K values compared well with values from slug tests, but were one to two orders of magnitude greater than those obtained from laboratory consolidation, falling head, centrifuge and triaxial tests. These data showed that this method can be used to determine in situ vertical K over short distances.